I think Robin has covered a lot of stuff really well. I will add a few more thoughts from my experience, taking Robin’s lead by replying to all of the points in one post.
Reply to #8
>(i) The spectrum gets wider (more height in Y-axis) as it goes down the sensor
>(ii) As well as this, the overall integrated flux gets bigger – an increase of around 50% from the two extremes explored here.
>(iii) However, the spectrum is not so good in the larger-flux
I wonder if these are all related. As I understand it you are moving the spectrum down the slit, taking a series of images. I am not sure you would expect the same quality of spectrum near the top of the slit, in the middle, and at the bottom of the slit. I suspect, but could be wrong, that the narrow vertical height of the spectrum might be due to some of the light being cut off by either missing the mirror or otherwise not following a path that successfully reaches the sensor. Hence, you see more flux with a wider height, which does not look as good, because this is how the spectrum naturally looks when all the light successfully passes through the spectrograph. I remember looking at this with my Lhires III. I also experimented with different mirror positions and at first thought great I am getting a nice tight spectrum. Only later did I look at the throughput and then realise I was getting less photons.
If you can get a nice spectrum in narrow vertical height with the same signal, then that is of course ideal. However, a spectrum spread in height of 10 or even 20 pixels is not necessarily a major problem. The geometric corrections should sort out the spectrum lines so they are nice and vertical, so you won’t loose much in resolution when the signal from different rows is combined. Also spreading the signal over a few pixels can be a good thing as then you have not got the problem of a single bad pixel or cosmic ray hit ruining a region of the spectrum.
Another way to investigate how the throughput changes with vertical position of the star on the slit is to take a sky spectrum at twilight, or possibly even daytime. It is like a flat, though not a flat you could use as it will be the solar spectrum. For this type of investigation the flat lamp may not be as good as it may not illuminate the slit evenly. The idea is by looking at the change in intensity with height, it will give you an idea of where to position your star for maximum throughput. Remembering the stellar spectrum does have vertical height, so this is not perfect, but it will give you a good feel for where maximum light passing through the slit to reach the CCD. Of course you can do this by positioning the star at different positions on the slit, but this gives a quick way to look at the whole slit.
On another topic related topic. I always try to position the star at about the same position in vertical height on the spectrograph CCD, as this minimises differences between spectra. Even though a flat field should remove most variations, I do everything I can to make create a repeatable accurate spectrum. When carrying out response or flux calibration, this means the light from both the reference and target stars following the same path through the spectrograph, and illuminate the same CCD pixels.
I found the grating holding was pinching my grating a little too tightly, and this was distorting the grating and so the spectrum. I loosed the holder, and then re-tightened it so the grating was held firmly in place. I saw a noticeable improvement in my spectrum, but it still had a vertical height of 10-20 pixels.